1

in the model y=B0+b1x1+bpxp+e, p=23, n=206 Sum square Reg =400, SS TOT= 1000 assume B1=B2=0 and deleting the predictor with these slopes ereased of the model.

Sum square= 100, SStot=1000 find F statistic.

I know this formula:

(SSreg1-SSreg2)/(p2-p1)/(ssReg/n-p2)

however the problem is I only have one P what should I do? or is it any other way?

Someoneii Knows
  • 31

1 Answers1

0

By setting $\beta_1=0$ and $\beta_2=0$, you are removing two of your $p=23$ variables. Therefore, your $p_1 = 23-2=21$.

By the way, there is a mistake in your formula.

$$F=\dfrac{\dfrac{RSS_{1}-RSS_{2}}{p_2-p_1}}{\dfrac{RSS_{2}}{n-p_2}}$$

Model $2$ has all $p_2=23$ parameters, while model $1$ only has $p_1 = 21$ parameters.

You might be interested in reading a good answer on the partial F-test, which is a particular type of chunk test.

Dave
  • 62,186